System and method for improved character recognition in distorted images

Abstract

Systems and methods are provided for recognizing characters within a distorted image. According to a one aspect, a method for recognizing one or more characters within a distorted image includes rendering one or more imitation images, the imitation images including simulations of the distorted image, applying one or more distortion models to the imitation images, thereby generating distorted imitation images, comparing the distorted imitation images with the distorted image in order to compute similarities between the distorted imitation images and the distorted image, and identifying the characters based on the best similarity. According to other aspects, the systems and methods can be configured to provide recognition of other distorted data types and elements.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS[0001]The present application claims the benefit of U.S. Patent Application Ser. No. 61 / 433,544, filed Jan. 18, 2011 which is hereby incorporated by reference in its entirety.TECHNICAL FIELD OF THE INVENTION[0002]This patent application relates generally to the field of character recognition and, in particular, to statistical methods for recognizing characters in distorted images.BACKGROUND OF THE INVENTION[0003]Optical character recognition (OCR) systems are employed in many fields and settings to identify characters within one or more images. For example, optical character recognition is commonly implemented for license plate recognition.[0004]While known OCR systems are often capable of generating accurate recognition results when the characters being identified appear prominently and clearly within the images being processed, such systems are far less accurate when processing images of sub-optimal quality. For instance, it can be appreciate...

AI Technical Summary

Benefits of technology

The patent text describes a system and method for recognizing characters in a distorted image using a computing device. The system includes a processor that receives a distorted image and a plurality of simulated images, each simulated using a different distortion model. The processor applies the distortion models to the simulated images and compares them with the distorted image to identify the characters. The system can also include a plurality of imitation data elements, each containing a unique simulation of the distorted image, and a similarity measure between the distorted image and the imitation data elements. The technical effects of the patent text include improved accuracy in identifying characters in distorted images and improved efficiency in processing large amounts of data.

Problems solved by technology

While known OCR systems are often capable of generating accurate recognition results when the characters being identified appear prominently and clearly within the images being processed, such systems are far less accurate when processing images of sub-optimal quality.

As such, existing OCR systems are frequently faced with distorted images that such systems are unable to process accurately.

In such situations, these systems either a) process such images inaccurately or b) abstain from processing such images (thereby requiring either a human operator to perform manual recognition of the image, or else not performing any recognition of the image).

Images